Application of convolutional neural networks in the diagnosis of schizophrenia

Objective To explore the program of convolutional neural networks for the diagnosis of schizophrenia and evaluate its effects. Methods Using the convolutional neural network, the training model was trained in the lead data of 138 normal people and 183 schizophrenic patients, and the model was validated by 20-fold cross-validation. Results The true positive rate of schizophrenia prediction using the convolutional neural network training model was 0.749, the false positive rate was 0.275, and the accuracy was 0.738. Conclusion This model can achieve a strong diagnostic ability for patients with schizophrenia.Therefore, convolutional neural network for the diagnosis of schizophrenia will become an important research direction in the future. Key words: Schizophrenia; Convolutional neural network; Electroencephalogram; Diagnosis